Re: Comments on FHS testsuite run

[Santiago Vila <sanvila@unex.es>]

On Wed, 25 Oct 2000, Wichert Akkerman wrote:

> Previously Santiago Vila wrote:
> > In either case, if by "infraestructure" you mean one or two empty
> > directories, I would say foreign apps which do not install rightly
> > because of these directories not being preset are broken, and we have
> > no need to support broken software.
> 
> I disagree. ISVs will write software for machines with the full expectation
> that they are LSB and FHS compliant. (At the moment we have the LDPS
> which says FHS compliant already). FHS clearly mentions that certain
> directories where certain data is supposed to be kept, and ISVs will
> expect those to exist. 

They will only expect those to exist if they interpret the standard in
the wrong sense, as the compliance-checker test currently does.

> We should not expect third-party software to check if all
> directories the FHS lists are present: [...]

Of course not, that's why there are mandatory directories and
non-mandatory ones. What we can expect is that they check that
non-mandatory directories listed in FHS are present so thay they are
created by the third party software in case they aren't.

In fact, we would be actually *helping* them to be fully standard
compliant if we do not create empty directories when they are not
really required by the standard.

This discussion started when a certain test said Debian did not follow
the standard in a number of places. I have yet to see where the
standard says those opt directories have to exist.

Thanks.